A Gcm Comparison of Plio-Pleistocene Interglacial-Glacial Periods in Relation to Lake El’gygytgyn, Ne Arctic Russia

Introduction Conclusions References Tables Figures

. Past warm periods known as Interglacials over the past 2.8 million years provide a means of studying global changes during a climate warmer than today giving us possible outcomes of trends seen in the modern world.
In 2009, a team of scientists from the United States, Germany, Russia and Austria drilled a 355 meter (1,165 ft.) sediment core from an 11 mile wide impact crater lake named "Lake El'gygytgyn" (alternatively, Lake "E"), in northeast Siberia. The recovered core contained the longest Arctic terrestrial record, extending back ~ 3.5 million years.
The sediment core revealed evidence for exceptionally warm periods in the Arctic; each lasting tens of thousands of years. These warm periods are marked by relatively large negative excursions of δ 18 O and are previously seen in ocean sediment cores. It has been shown that Marine Isotope Stage(s) 5e, 11c and 31 were some of the warmest interglacials in the Pleistocene Arctic (Melles et al., 2012) and are of important interest as 2 they can be considered an analogue for a warmer, future Arctic that our climate may be heading toward.
An interglacial period (alternatively called an interstadial) is a period in geological time that is marked by warmer than average global temperatures that last thousands of years (fig 1.1). Evidence of such periods lie within many geological records such as deep-sea sediment cores, ice cores and speleothem analysis and are marked by large negative oxygen isotope (δ 18 O) excursions in the oxygen isotope record obtained from the composition of foraminiferal carbonate (Shackleton, 1967). The blue and green curves show temperature from two sites in Anatarctica derived from Deuterium measurements (δD) on ice cores (Augustin et al., 2004;Petit et al., 1999) (Kitoh & Murakami, 2002).

Marine Isotope Stage 5e
Interglacial 5e is one of the warmest interglacials of the Pleistocene and lasted roughly ~12-10 kyr (130 to 116 kyr). High obliquity, eccentricity and precession allowed for very warm summer orbit and insolation intensity maximum at around 127 kyr.
During the Last Interglacial (LIG), a warmer climate throughout the Arctic possibly caused a size reduction of the Greenland Ice Sheet (GIS). Studies involving Sr -Nd -Pb isotope ratios of silt-sized sediment discharged from southern Greenland suggest that no single southern Greenland geologic terrain was completely deglaciated during the LIG, however, greater southern GIS retreat was evident (Colville et al., 2011). Additional analysis of MIS 5e by Yin & Berger (2011) -Bliesner, 2006). Additional warming during this period supports the notion of a significantly reduced Greenland Ice Sheet.

Marine Isotope Stage 11c
Interglacial 11c is another exceptionally warm interglacial that lasted from 428 to 383 kyr (~45 kyr). Sediment records containing information on MIS 11 are very uncommon in the Arctic and temperature data are inadequate (Miller et al., 2010).
Unlike the other interglacials, MIS 11c was remarkably long with two insolation maxima anomalies at ~ 409 kyr and 423 kyr, creating extensive warmth throughout the Arctic.
Similar to MIS 5e, there is evidence that the GIS may have been reduced (Raymo & Mitrovica, 2012;Willerslev et al., 2007) with lush boreal forest covering most of southern Greenland (de Vernal & Hillaire-Marcel, 2008). Particularly warm conditions are also suggested by pollen records analyzed from Lake Biwa (Tarasov et al., 2011).
Likewise, a similar study done with a record from Lake Baikal also shows warmer than modern temperatures with a "conifer optimum" suggesting not only warmer conditions but also less continentality and higher sea levels than present (Prokopenko et al., 2010 1.2). Lake El'gygytgyn resides in an area of the western arctic known as Beringia, which is bounded by the 140° E meridian to the west and the Alaska/Canada border to the East, the 76° N parallel to the North, and the 50° N parallel to the South (Mock et al., 1998). Beringia is separated into two distinct geographical regions.
Western Beringia, where the lake is located, is west of the Bering/Chukchi Sea and 6 Eastern Beringia is east of the Bering Sea and incorporates all of Alaska and the Yukon.
January monthly mean temperatures range from -47 to -5 °C in Western Beringia to -30 to 0 °C in Eastern Beringia (Mock et al., 1998). Averaged July monthly temperatures tend to increase with decreasing latitude, with values ranging from 0 to 16 °C (Mock et al., 1998). For most of Beringia, precipitation is at maximum during mid-to late-summer with averaged July precipitation amounts varying from 50-100 mm (NCDC, 1999).
Combined, both Western and Eastern Beringia have July averaged monthly precipitation in the range of 25 to 225 mm including both Western and Eastern Beringia.
Today, mean annual air temperature at the lake is -10.4 ± 1.1 °C with daily average temperatures during the summer (JJA) ranging from 0 to 12 °C (Nolan & Brigham-Grette, 2006). Extremes in 2002 ranged from -40 °C in winter to as high as +26 °C in the summer with occasional mid-winter warming approaching 0 °C (Nolan & Brigham-Grette, 2006). Precipitation amounts at the lake are rather small indicating a dry environment typical of the Arctic (<300 mm year -1 ). Weather stations implemented around the lake in 2002 recorded 70 mm of rainfall all between mid-May and late-September with transient snowfall greater than 5 cm beginning in mid-July and lasting the rest of the summer and 178 mm of precipitation from the end of summer 2001 to end of summer 2002 (Nolan & Brigham-Grette, 2006).

Research Statement
The primary goal of this investigation is to study Arctic climate variability and sensitivity to the exceptionally warm interglacials during the past 2 million years (Pleistocene) and correlate the modeling data with the Lake El'gygytgyn multiproxy analysis. By studying the interglacial and glacial climates, Arctic variability can be assessed and quantified with an aim toward studying the teleconnections associated with them. This is especially important as the Earth continues to warm due to anthropogenic emission of Greenhouse Gas (GHG). The work discussed in this thesis will advance on the work already done on MIS 5e, 11c and 31. Such advances on the original work include 3 simulations with high-resolution interactive vegetation to show 1) biome regimes and vegetation feedbacks, 2) the sensitivity of the circum-Arctic to the Greenland Ice Sheet, 3) the sensitivity of the lake region to the reduced Arctic sea ice owing to the intrusion of warm north Pacific waters into the Arctic Ocean and 4) circum-Arctic and Lake El'gygytgyn sensitivity to major northern hemisphere ice sheets. 8 The document presented here is a progression from the description of the various tools and methods used, while also describing the model boundary conditions and details of the model simulations (Chapter 2). This is followed by a presentation of the results from each model simulation discussed in detail (Chapter 3). Discussion of simulations and overall climatic patterns are explained in (Chapter 4). Chapter 5 discusses the glacial paleo-arctic environment before the interglacial transition. Conclusion of analysis and research statement summary will be displayed in (Chapter 6). All figures and tables embedded throughout the text will be labeled and noted in the table of contents. 9 CHAPTER 2

Mathworks MATLAB R2011b
Mathworks MATLAB is a high-level language and interactive environment for matrix-based numerical computation. All post-processing and analysis was conducted using computer-based programs (alternatively called scripts) executed in MATLAB.
MATLAB provides both high and low-end access to NetCDF (Network Common Data Form) files, which are produced by the model, allowing read and writes capabilities from the MATLAB interface to the NetCDF format. MATLAB also contains low-level access to common NetCDF functions in order to access NetCDF libraries. The ability for MATLAB to access such libraries with large amounts of data makes it an ideal program to visualize and compute model output data.

Global Climate Model: GENESIS Version 3.0
All global climate simulations discussed in this thesis were performed using the Global ENvironmental and Ecological Simulation of Interactive Systems (GENESIS) Global Climate Model (GCM) version 3.0 (Thompson & Pollard, 1997). The GCM is written in Fortran and ran in parallel on a Silicon Graphics (SGI) computer running 10 years per day. GENESIS is an atmosphere, land-surface, ocean, snow, sea-ice, ice sheet and vegetation coupled model. Spectral resolution of the 3-D atmosphere GCM (AGCM) within GENESIS is T31 resolution (3.75° lat. x 3.75° long.) with 18 vertical levels (Thompson & Pollard, 1997). The AGCM is coupled to the land-surface model by a Land-Surface-Transfer scheme (LSX), which computes fluxes through the vegetation 10 model . In addition to a coupled AGCM and land-surface scheme, GENESIS allows for a 50-meter, mixed-layer, non-dynamical, slab ocean model that incorporates heat transfer, calculations of sea-surface temperatures (SST) and feedbacks operating between ocean surface and sea-ice. This version of the GCM has sensitivity to 2xCO 2 of 2.9 °C, without GHGs, vegetation or ice sheet feedbacks.
The sea-ice model is based upon the sea ice component used in Washington & Meehl (1996) in a fully coupled atmosphere, ocean and sea ice model. Sea ice dynamics are based upon the cavitating fluid method derived by (Flato & Hibler, 1992). The sea ice component is driven by momentum; heat and freshwater fluxes provided at the upper and lower ice boundaries from the atmospheric and oceanic model components. A Flux Coupler (Bryan et al., 1996) facilitates and manages the exchange of fluxes between the model components and conserves equations of heat, momentum and freshwater within the model climate system (Weatherly et al., 1996). Sea ice is able to drift in the model using the shear stress of the wind across the upper boundary of the ice.

AGCM Overview
The 3 thermal infrared radiation code. The solar radiation scheme of (Thompson et al., 1987) is 11 included, which combines all clouds into a single effective layer for solar computations and allows for effects of atmospheric aerosols.
Clouds in the GCM are parameterized similar to Slingo & Slingo (1991) for three different clouds types: stratus, anvil cirrus, and convective .
A constraint on stratus clouds is used when specific humidity is very low permitting sensible amounts of clouds to form over Polar Regions during winter . Clouds in the GCM are formed using a plume model similar to (Kreitzberg & Perkey, 1976) but does not include cloud microphysics. At each horizontal gridpoint and at each time step, a column model of subgrid rising thermals is solved including saturation and precipitation . From the previously solved equations, large-scale vertical fluxes, latent heat and precipitation can be inferred. Similarly, the planetary boundary layer (PBL) is calculated using the same model by initiating dry PBL thermals at the center of the lowest model layer (Morton, 1968;Telford, 1966). Surface (2° lat. x 2° long.) and AGCM (3.75° lat. x 3.75° long.) fields are coupled to one another by two different model schemes: bilinear interpolation (AGCM to surface) and forward area averaging (surface to AGCM) at each time step.

BIOME4 Vegetation model Overview
GENESIS GCM is fully coupled to the BIOME4 (Kaplan, 2003) interactive vegetation model that was developed from the BIOME3 model of (Haxeltine & Prentice, 1996). BIOME4 is a coupled carbon and water flux model that predicts equilibrium vegetation distribution, structure and biogeochemistry. Vegetation distributions take the form of 27 plant biomes including 12 plant functional types (PFTs) that represent broad, 12 physiologically distinct classes ranging from cusion-forbs to tropical rain forest trees (Kaplan, 2003). Each PFT is assigned limits in relation to climate, which would define whether or not the functional type exists within that grid cell. Identification of the biome in each grid cell is determined by the ranking of the PFTs, given by the model. The ranking is based on biogeochemical variables, such as leaf area index (LAI), the monthly mean climatology and mean annual soil moisture, which determine the appropriate biome.

Biome Type 1
Tropical evergreen forest 2 Tropical semi-deciduous forest 3 Tropical deciduous forest/woodland 4 Tropical xerophytic shrubland 5 Temperate xerophytic shrubland 6 Tropical grassland 7 Temperate grassland 8 Temperate conifer forest 9 Warm mixed forest 10 Cool mixed forest 11 Cool conifer forest 12 Cold mixed forest 13 Temperate deciduous forest 14 Evergreen taiga/montane forest 15 Deciduous taiga/montane forest 16 Tropical Savannah 17 Temperate broadleaved savanna 18 Open conifer woodland 19 Temperate sclerophyll woodland 20 Boreal Parkland 21 Steppe Tundra 22 Shrub Tundra 23 Dwarf shrub tundra 24 Prostrate shrub tundra 25 Cushion-forbs lichen and moss 26 Desert 27 Barren 28 Land ice In addition to control over Milankovitch parameters, GENESIS also allows Greenhouse gasses to be prescribed uniformly. GENESIS's namelist parameters allow changes in Carbon Dioxide (pCO 2 ), Methane (CH 4 ), Nitrous Oxide (N 2 O) and Chlorofluorocarbons (CFCs). Simulations of each interglacial were run with the proper GHG concentrations from the literature and orbital parameters from Berger's algorithm.

GENESIS Boundary Conditions
Boundary conditions in GENESIS were initiated on a 2° lat. x 2° long. surface grid (90 rows x 180 columns). Conditions upon startup are default values within each topography and surface input file. The input parameters read by the model include surface type, gravity-wave roughness, topography, vegetation, ocean-lake fraction, 16 atmospheric ozone distribution, soil texture and depth (Thompson and Pollard, 1995[guide]).
Surface topography (2 ° long. x 2 ° lat.) editing is done through data input files that are interconnected to the AGCM (T31 resolution). Default values in the topography files are derived from the U.S. Navy FNOC global elevation dataset at 10 min. resolution (Cuming & Hawkins, 1981;Kineman & Hastings, 1992) in Fortran I5 and are measured in mean sea level in meters. Surface files are coded with 1=land, 2=ice sheet and 3=ocean in Fortran A1 format. Ice sheet areas were superimposed using Cogley's 1° x 1° Global Hydrographic Dataset (Cogley, 1991;Pollard and Thompson, 1995[guide]).
Greenhouse Gasses are prescribed during initial startup using the model namelist parameters in the configuration file.

Paleoclimate Boundary Conditions
Topography during this study was changed remained largely unchanged, except for simulations of an ice free Greenland where exceptionally warm conditions in the Arctic during interglacials 31 and 11c (Elias & Matthews Jr., 2002;Melles et al., 2012;Raymo & Mitrovica, 2012) prevailed, and a change in Greenland topography and surface type is required if simulations are to be accurate. Removing Greenland's ice sheet requires changing the surface type and topography input files. Such an edit necessitates a change in elevation of Greenland's topography by +6 meters to simulate glacial isostatic adjustment (GIA) in each grid cell and a surface type change from ice to land. Similarly, edits of topography and sea level were also needed in paleoclimate simulations with large 17 Northern Hemisphere ice sheets. Such ice sheet data was extracted from ICE4G (Peltier, 1994) dataset of ice and water cover since the Last Glacial Maximum (LGM).
In simulations when vegetation is not interactive, and is prescribed rather than simulated, vegetation and biome distribution input files, similar to those of topography and surface, must be edited to the correct biome. Biome designations are labeled 1-12 based on (Dorman & Sellers, 1989) vegetation type and designate a single globally uniform vegetation type for all land points (Thompson and Pollard, 1995 [guide]).

Greenhouse Gas concentrations
Greenhouse Gas concentrations were prescribed uniformly. Since MIS 31 lies beyond the age of the oldest ice core record, atmospheric Carbon Dioxide (pCO 2 ) concentrations were prescribed from boron isotopic compositions of foraminifera shells (Honisch et al., 2009)  Greenhouse Gas forcing contributions (Table 1) were calculated using the IPCC simplified calculations for radiative forcing due to CO 2 , CH 4 , N 2 O, and halocarbons, the latter omitted from our experiments (Smithson, 2002). 18

Trace Gas
Simplified expression Radiative forcing, ∆F (Wm -2 )  (1998). (Table adapted from (Smithson, 2002)  To test the potential effect of a warmer Arctic Ocean with reduced sea ice on the Beringian interior, the slab ocean model component was modified by adding an additional 8 W m -2 of ocean heat convergence under sea ice, in addition to the 2 W m -2 used in modern control simulations (10 W m -2 total). This mimics the potential impact of a substantial enhancement of oceanic heat flux to the Arctic basin at times of high sea level (MIS 11c). The increased heat flux is based on a simple calculation assuming an extreme 3 Sv increase in Bering Strait throughflow and a 4 ºC temperature contrast between North Pacific and North Polar surface water. This should be considered a simplistic sensitivity test that should be constrained by future ocean modeling studies.

Additional information on facies interpretation
Sedimentation in Lake El'gygytgyn is highly variable. Distinct lithofacies of the pelagic sediment record were defined based on the physical characteristics of the sediments, including color, particle size, and the presence or absence of various sedimentary structures as visually observed in the split core halves and high-resolution radiographs (100 µm resolution) obtained using an ITRAX core scanner (Fig. S2). Fine-scale details of characteristic type-sections were further investigated using thin-sections prepared from epoxy-impregnated sediment slabs. Highresolution digital images and backscattered Scanning Electron Microscope (SEM) images were used to evaluate the thin sections.
Facies A is defined by the presence of fine clastic laminations (<5mm in thickness; average is ~0.2 mm). Sediments of Facies A are predominantly dark gray to black in color. Laminations have distinct lower bounding surfaces and grade upwards from silt to clay before repeating.  (Berger, 1978)  and precipitation and correlate the data to pollen proxy analysis. Orbital and GHG values are estimated for 127 kyr; peak warmth during MIS 5e.

Experimental Run -Marine Isotope (MIS) 11c, 409 kyr
In this section, there will be three different simulations to test the sensitivity of the lake region during MIS 11c. The first simulation will be done with default boundary conditions, including a Greenland Ice Sheet and will be referred to as MIS11GIS. The second simulation will test the sensitivity of the Arctic to an ice-free Greenland, hereafter known as MIS11NG. The scientific literature shows that during interglacial 11c, the Greenland Ice Sheet was significantly reduced and warm boreal forests (spruce, alder, etc.) covered parts of the island (Raymo & Mitrovica, 2012;Willerslev et al., 2007).
Consequently, the GIS was removed and topography of Greenland was corrected for glacial isostatic adjustment (GIA) within the appropriate model topography files. The final sensitivity experiment involved an increase in sub-oceanic surface heat flux from 2 Wm 2 in our modern control, to 10 Wm 2 (additional +8 Wm 2 ) to test the Beringian sensitivity to an ice-free Arctic Ocean. Today, the Bering Strait is limited to ~ 50 m in depth with a northward transport of ~ 0.8 Sv (Woodgate, et al., 2010). The increase heat flux assumes an extreme 3 Sverdrup (Sv) increase in Bering Strait throughflow and a 4 °C temperature contrast between North Pacific and North Polar surface water (Melles et al., 2012, supplemental). The additional heat flux convergence was used to simulate increases in energy flux through a wider Bering Strait during times of higher sea level.
Using BIOME4, comparison of Arctic vegetation within the Beringian region can be analyzed in order to compare model and pollen proxy data that were collected from Lake 21 El'gygytgyn. Furthermore, fixed vegetation studies using BIOME4 will isolate and quantify the forcing effect of vegetation on surface temperatures around the lake region.
Concentrations of GHGs will be prescribed as: 285 ppm pCO 2 , 713 ppb v CH 4 , and 285 ppb v N 2 O. In terms of orbital parameters, Obliquity and Eccentricity will be set to a value of 23.78°, 0.019322 (Berger, 1978) respectively and omega as 276.67 converted to 263.33 (eq. 1)(see table 2.3).

Experimental Run -Marine Isotope Stage (MIS) 31, 1072 kyr
This period, in addition to MIS 11c, was also speculated to be too warm for a Greenland Ice sheet to exist (Melles et al., 2012). Therefore, model runs with and without a Greenland ice sheet (including glacial isostatic adjustment) were executed to show sensitivity and forcing feedback for these scenarios.
Concentrations of GHGs will be prescribed as: 325 ppm pCO 2 , 800 ppm v CH 4 , and 288 ppb v N 2 O. Orbits from Yin and Berger (2011) show a very warm orbit with Obliquity and Eccentricity of 23.89° and 0.05597, respectively. Precessional value is rather large, with a value of 289.79 after conversion to model specific value (eq. 1).

Experimental Analysis
In order to test the sensitivity of the Arctic, especially the Western Arctic to changing boundary conditions, data, such as temperature, precipitation and vegetation, were plotted and compared to control runs (Pre-Industrial and Modern control). The goal of this study is to observe the Arctic's climatic and terrestrial response to high levels of greenhouse gasses and warm orbits that coincide with the interglacial periods. Such 22 responses being studied are the effects on temperature, precipitation and vegetation in and around the lake region. Moreover, analyses of atmospheric properties such as sea level pressure and geopotential heights were analyzed to show pressure anomalies that may be linked to changes in topography and ice sheets in the circum-arctic. Using these data, comparisons of model output temperatures and precipitation relative to Pre-Industrial and Modern control runs can be studied. More importantly, pollen analysis done on the lake core (Melles, Brigham-Grette et al., 2012) can be validated by analyzing surface temperatures and precipitation whereas also validating plant assemblages by using vegetation output from the BIOME4 interactive vegetation component of GENESIS, or vice-versa. Possible changes in atmospheric circulation, temperature and precipitation due to regional changes in topography and ice sheets will also be considered and associated to control scenarios.

Model Output Post-Processing
GENESIS surface-model history files (LSX) contain 54 variables in monthly mean data sets. Analysis using these history files will focus mainly on 2-meter surface temperature and precipitation. Likewise, AGCM history files contain 38 variables in monthly data sets. The variables used here during the study were 500 hPa geopotential heights, surface and sea-level pressure and insolation at top of the atmosphere (TOA).
Simulations of the specific time periods were ran for 30 to 40 years to ensure model climate equilibration with initial conditions and a 50-meter slab ocean. For analysis, the last 10 years of data (20-30; 30-40) was extracted and averaged over a 180 x 90 x 12 grid.
This grid represents 180 degrees of longitude, 90 degrees of latitude containing 12 23 months (one year) of data. On this grid, averaged monthly or yearly data can be plotted on a map projection allowing it to be visually attractive for publishing and easily examined for data analysis.

BIOME4 Output Processing
Biome vegetation output was analyzed by accessing the last year of vegetation in the equilibrated run. Additionally, averaging biome data becomes a programming challenge. This is due to the arrangement of the 28 biomes and the fact that, for example,  To further test the validity of the GCM temperatures, a comparison was made with National Center for Environmental Prediction (NCEP) Reanalysis data. The difference indicates that GENESIS is only + 0.5 °C warmer than the modern reanalysis data in the lake region signifying relatively reliable results when doing calculations with 26 July surface temperatures (fig 3.1). Yet, GENESIS presents a warm bias over Greenland and parts of Northeastern Canada, and a cold bias in central, interior Russia compared to NCEP data.

Precipitation
Control simulation of Mean Annual Precipitation (PANN) (fig 3.1 B)  with modern observations and field data of modern, Arctic vegetation in the region (Kolosova, 1980;Viereck & Little Jr, 1975). With this said, I hypothesize that the vegetation is not in full balance with the environment suggesting it is still transitioning into equilibrium. 29 . Y-axis is latitude and Xaxis is months (1 -12; Jan. -Dec.). Red star denotes location of Lake El'gygytgyn. Please note Modern simulated vegetation around the lake is not in equilibrium and suggest conifer forests instead of shrub tundra, biome #22.

Temperature
Simulations of pre-industrial 2-m MAAT and MTWM at Lake El'gygytgyn are -12 and 10.3 °C, respectively. This is to be expected, as pre-industrial GHG levels are lower than those of present day. Thus, lake regional annual air and July temperatures are -3°C and -1.7°C cooler than those of the modern simulations, respectively (fig 3.4 D, C).
Similarly, summer temperatures are cooler as well, on the order of -2.2 °C cooler (8°C) (fig 3.3 A) than modern temperatures. Although Earth's orbit, specifically obliquity, has not changed in 120 years, temperatures are still cooler than modern temperatures. This is largely due to the fact that low CO 2 emissions during this period attenuate the effect of yearlong radiation being transferred from the atmosphere back to the surface. GHG radiative forcing from a combination of CO 2 , CH 4 , and N 2 O atmospheric mixing ratios determined from the literature indicates a -1.8 Wm -2 change relative to modern GHG radiative forcing. CO 2 radiative forcing contribution alone is the largest contributor to the decrease in forcing feedback (-1.3 Wm -2 ), all contributing to the cooler surface temperatures.

Precipitation
Generally, PANN values in the pre-industrial simulation showed slightly lower values than that of our modern precipitation values. Annual precipitation was around 438 mm year -1 (+122 mm year -1 relative to obs.) (fig 3.3 B) indicating slightly drier conditions in the lake region coinciding with a cooler, pre-anthropogenic warming environment. The same can be said for precipitation amounts in Northwest Yukon and 32 North Pacific where amounts prior to this run were +200 mm year -1 higher. Mean winter precipitation range was about 25 mm month -1 , while mean summer precipitation was 43 mm month -1 , indicating -1 and -20 mm month -1 less precipitation relative to modern control, respectively. Most of the circum-Arctic experiences drier conditions during the seasons, with wetter conditions prevailing in the modern runs.

Vegetation Distribution
Though     Somewhat drier conditions prevail in interior Siberia and may be linked to lack of moisture and enhanced continentality.

Vegetation Distribution
Most of Alaska is covered with evergreen taiga forest with deciduous toward the north coast. Lake El'gygytgyn is in a transition zone with dominant shrub tundra to the east and deciduous forest to the west (fig 3.9)      Red star denotes location of Lake El'gygytgyn. Vegetation correlates well with increase of trees and shrubs in the Lake El'gygytgyn multiproxy record during peak insolation anomalies. 42

Temperature
Overall warming of the Beringian interior was +5 (±1) °C relative to modern temperatures. Mean annual summer and July temperatures during interglacial 5e show 11 and 14.5 °C, respectively (fig 3.10 A, B). summer warming over the GIS reflects +5 °C warmer than pre-industrial and only roughly +1 °C warmer than modern simulations.

Precipitation
Mean annual precipitation during MIS 5e is about 401 mm year -1 (fig 3.13 E), which is -74 and -37 mm year -1 less than modern and pre-industrial levels, respectively.
Overall, similar precipitation patterns are seen over the Arctic between MIS 5e and the pre-industrial control scenario. 43

Vegetation Distribution
Most of Beringia and the lake region is covered by deciduous taiga (fig 3.14) and

Marine Isotope Stage 11c (409 kyr)
Marine Isotope Stage 11c is a long interglacial compared to the other interglacials in this study. We assume an ice-free Greenland in our MIS 11c simulations, with the ice sheet removed and replace with isostatically equilibrated (ice-free) land elevations.
Additional experiments involving sea-ice extent will also be mentioned with the results outlined.

Temperature
Contribution of summer insolation forcing during this period ranges from +45 - In similar simulations performed with a modern Greenland Ice Sheet (GIS), temperature difference with and without a modern GIS was negligible, as the loss of the ice sheet only created July warming of ~0.3 °C around the lake. Warming, albeit slight, was present when comparing geopotential height anomalies around the lake. Anomalies of +4 -10 meters indicate warming of the column of air above the lake, with negative height anomalies to the west of the lake.
The warmer climate across the Arctic and reduced GIS was thought to have increased sea levels by as much as >11 meters (Raymo & Mitrovica, 2012) with little sea 50 ice extent. In order to test high sea levels and an ice-free Arctic Ocean around Lake El'gygytgyn, increased subsurface heat flux convergence from 2 Wm -2 to 10 Wm -2 was initiated. The resulting reductions in sea ice and warmer Arctic SST's produced negligible warming in the Beringian interior around the lake (< 0.7 °C). Interestingly, boreal forest biome forcing on surface temperatures was quantified around the lake region presenting a net cooling of -2 °C rather than warming.

Precipitation
Precipitation amounts at the lake during MIS11GIS are very similar to modern precipitation amounts of 475 mm year -1 (fig 3.17 E). Also, MIS11NG exhibits exact precipitation amounts as our pre-industrial control run (~438 mm year -1 ). Rainfall conditions directly in the Arctic Ocean basin are very dry, ~200 mm year -1 , which is year -1 ).

Vegetation Distribution
The Lake El'gygytgyn region during MIS 11c is on the border of evergreen taiga and shrub tundra biomes (fig 3.18 G). Most of interior Siberia is deciduous forest and temperate grassland, similar to MIS 5e and 1. Interior Alaska and Yukon are mostly evergreen taiga and some deciduous forest toward the northern shore of Alaska, with 51 incorporated sporadic shrub tundra mixed in. With the loss of the GIS, Greenland is now predominantly shrub tundra with dwarf shrub tundra along the northern shore.
Vegetation limits, such as tree lines, are slightly changed during our simulations with increased heat flux and a warmer, open Arctic Ocean. Evergreen forests around the lake region and in Alaska extend poleward toward the coast, and deciduous forest is replaced by shrub tundra on the northern coast of Alaska (fig 3.18 H). Evergreen forest in the Yukon continues to be dominant with an eastward migration of the tree line taking over some grassland, as Greenland remains unchanged. 52

Temperature
A very warm orbit with high obliquity, eccentricity and precession aligning perihelion with boreal summer allows insolation anomalies to be > 50 Wm -2 at the surface and + 60 -80 W m -2 at the top of the atmosphere (fig 3.21 F). Average summer temperatures around the lake are about +1.6 °C (fig 3.20 C) warmer than modern and +3.6 °C (fig 3.19 A) warmer than pre-industrial. CO 2 forcing contributions of +0.80 Wm -2 relative to pre-industrial values, permit July temperatures to exceed +5 and +3.5 °C warmer than pre-industrial and modern temperatures, respectively (fig 3.19 B; 3.20 D).
Most summer warming is seen over Greenland and interior Siberia with temperatures over an ice-free Greenland of +15 -17 °C and interior Siberia, with temperatures +6 -8 °C warmer relative to pre-industrial and modern temperatures.
July average temperatures are in similar agreement relative to modern and pre-industrial control mean summer temperatures with overall >15 °C warming over Greenland during this period.

Precipitation
Overall precipitation in the Arctic during interglacial 31 is ~ 438 mm year -1 , similar to that of interglacial 11c (fig 3.21 E). However, summer precipitation is similar to modern values; about 65 mm month -1 indicating more water vapor in the air possibly correlated with increased temperatures. 57

Vegetation Distribution
Vegetation distribution is similar to most of the interglacials described here. Most of the Alaskan interior is dominated by evergreen taiga forest with only a few areas of shrub tundra on the coasts. Lake El'gygytgyn is dominated by deciduous taiga with evergreen dominating toward the eastern coast (fig 3.21). Most of interior Siberia shifted from once being predominantly deciduous forest to now being only half deciduous forest and an expanding area of temperate grasslands. Without a GIS dominating interior Greenland, the landscape has shifted from tundra in MIS 11c to mostly evergreen forest.
Interior Yukon remains the similar to other interglacials with a mix of temperate grassland and evergreen forests. 58   (Miller et al., 2010). Powerful insolation forcing at these latitudes permits July maximum temperatures to exceed both pre-industrial and modern temperatures by at least >3 °C which is in agreement with the previous study. The 2 -4 °C warming in Siberia and western Beringia in our results has been shown by simulations with a model without vegetation feedbacks and has been linked to strong summer insolation anomalies (Otto-Bliesner, 2006). Anomalous insolation forcing was shown between 130 and 127 kyr during the summer season, the maxima at which our GCM was simulated. Moreover, the exceptional summer warming compared to other interglacials was thought to have caused a reduction in the Greenland Ice Sheet adding 1.6 to 2.2 m of equivalent sea level rise (Colville et al., 2011). A more recent study conducted by the North Greenland Eemian Ice Driling Project (NEEM) confirmed that the thickness of the Northwest sector of the GIS decreased by 400 ± 250 meters reaching surface elevations of 130 ± 300 meters lower than present (Dahl-Jensen et al., 2013). This indicates that our simulations of MIS 5e with a near-modern GIS are a good approximation for this period. Increased warmth allows almost a full replacement of shrub tundra with deciduous forest in and around the lake region. Pollen analysis during this period show tree species of birch, alder, pine and spruce (Melles et al., 2012). However, multiproxy studies of MIS 5e show a change in MTWM of only +2 °C warming at the lake compared to modern temperatures (Melles et al., 2012). I conclude that a warm summer orbit with only moderate GHG concentrations does account for exceptionally warm temperatures in Beringia however, the particularly 64 muted response in the Lake El'gygtgyn proxy record to summer insolation forcing cannot be fully explained (fig 4.1 I).
Simulations of 11c exhibit another very warm interglacial in the Arctic around the lake with MTWM maxima approaching +2.2 °C warmer than pre-industrial temperatures.
Similarly to MIS 5e and 1, peak warmth coincides with perihelion during boreal summer however, a low eccentricity and obliquity attenuates the effects of precession relative to 5e and 1, making summer less intense, although longer in duration. This noticeable warmth is an obvious outcome of low to non-existent snow-ice albedo effect contributing to extreme warmth. Under the assumption sea level has risen due to ice sheet melt, increasing heat flux convergence under sea ice in the Arctic Ocean from 2 to 10 W m -2 allowed us to test the hypothesis whether lessened sea ice and increased SSTs warmed the lake region by using a simplistic sensitivity test based on a modest calculation of latitude interglacial forcing on terrestrial biome distribution is evident in our simulation by a poleward advance of evergreen needle-leaf forest during the interglacial around the lake which is in agreement with palynological analysis of tree species in the lake area 65 (Melles et al., 2012). Analysis suggests forest-tundra and northern larch taiga environments with dark coniferous forests dominant of spruce, pine, birch, alder and larch controlled the lake region (Melles et al., 2012). Enhanced solar anomalies drove interior locales to warm allowing boreal forests to thrive. Surface warming as a result of increased low albedo needle-leaf forests accounts for some of the warming seen at the surface during this period. However, isolated forcing feedback of increased evergreen, terrestrial forest provides a net cooling effect during the summers and slight net warming effect during early fall (Sep. -Nov.; +0.3 °C).
A deglaciated Greenland has been shown to have regional effects on SSTs and sea-ice conditions, however warming of the circum-Arctic has been shown to be minimal (Koenig et al., 2012). This was demonstrated in our simulations by isolating the effects associated with the loss of the GIS leading to warming around the lake of only +0.3 °C.
Analysis of 500 hPa geopotential height anomalies exhibit ridging (

Introduction
An additional sensitivity test of Lake El'gygytgyn to changing boundary conditions associated with the buildup of major northern hemisphere ice sheets was also simulated and related to pollen analysis at ~2.7 Ma in the lake core. Such a substantial

Method and Experiment Set-up
Two simulations were run using the GCM described earlier in this thesis (GENESIS GCM, v. 3.0) with (3HL116K) and without (3NG116K) Northern Hemisphere ice sheets. In both cases, the GCM was run to equilibrium with averages calculated from the last 10 years of the model's history files. The first simulation used ice-free Northern Hemispheric climate conditions, while the second simulation adds the Greenland, Laurentide and Fennoscandian ice sheets, based on the LGM ice volume from ICE 4G (Peltier, 1994;Brigham-Grette et al., 2013) including a decreased sea level. This 70 simple sensitivity test is used to show the effect of large Northern Hemisphere ice sheets on Arctic climate.

Temperature
Mean Temperature of the Coldest Month (MTCM; Jan.) around Lake El'gygytgyn was simulated to be -40 °C with July temperatures about 3 °C (-5 °C relative to modern temperatures). These temperatures compare favorably with proxy reconstructions after 2.7 Ma and pollen reconstructions of the cool periods between interglacials (Brigham- Grette et al., 2013;Melles et al., 2012). Mean annual temperatures in the circum-Arctic decrease 5 to 25 °C in response to the increase of large ice sheets relative to the experimental run without Northern Hemispheric ice sheets.

Precipitation
Preliminary GCM analysis of mean annual precipitation (PANN) shows that most of the circum-Arctic becomes very arid with more than 150 mm year -1 decrease in

Discussion
Arctic aridification and temperature change can be linked to mechanical atmospheric forcing associated with large northern hemisphere ice sheets. Exceptionally large temperature decreases are thought to be associated with albedo-enhanced cooling from large ice sheets reflecting solar radiation back to the atmosphere. Likewise, enhanced cooling in the Arctic and expanded sea-ice cover contributed to circum-Arctic aridification (> 150 mm year -1 ).
Comparable studies (Bromwich et al., 2004) using regional climate models to quantify mechanical forcing of large northern hemisphere ice sheets show important effects on mid-tropospheric westerly flow. The presence of a very large Laurentide ice sheets splits the jet stream into two branches: a northern most, polar jet and a southern branch ( fig 5.1 A, B, C). Due to this split flow around the ice sheet, during January, surface cyclones tend to flow along the periphery of the Laurentide Ice sheet due to a very strong high-pressure system that forms over North America (fig 5.3, A). Due to a strong mid-level trough that forms on the south coast of western Beringia (fig 5.3, B), storms are frequent along the southern coast of Alaska and Beringia (Bromwich et al., 2004). During the summer (July), the jet stream is positioned directly over the ice sheet 72 allowing increased frequency of surface cyclones to migrate directly over the ice sheet dropping 42% of annual precipitation (Bromwich et al., 2004). This can be attributed to a large trough centered over southwestern North America (fig 5.3, C), allowing the storm track to push storms further south over this region and North America. Additionally, Beringia is encased in a very strong high-pressure system (fig 5.3, D), presumably limiting precipitation in Beringia and at the lake. This strong high-pressure system seems to be related to a considerable strengthening of the Siberian high. It is important to note that the strengthening of the Siberian high is seen when we have large Northern Hemisphere ice sheets. It can be concluded that mechanical forcing of northern hemisphere ice sheets led to aridification of the Arctic due to changes in the dominant storm track patterns. Even though these results are not definitive, the results suggest that the presence of large Northern Hemisphere ice sheets contributed to changes in synoptic weather patterns leading to aridification of Lake El'gygytgyn and the change of boreal/evergreen forest around the lake to shrub tundra, lichen and mosses. which created the greatest summer warmth out of all interglacials, produces high-77 intensity summer insolation of >50 Wm -2 at the surface and ~ -.98 --1.89 Wm -2 of greenhouse gas forcing, relative to modern values (Melles et al., 2012). MIS 1 is an exception with lower CO 2 around the time of peak Holocene warmth producing -0.44 Wm -2 less radiative forcing relative to pre-industrial levels (Melles et al., 2012).
Extreme warmth and changes in greenhouse gasses shifted vegetation from mostly tundra with small shrubs as we see the Arctic today to thick, lush evergreen and boreal forest. Due to the extreme warmth, wetter conditions prevailed during the superinterglacials allowing biomes to thrive and increase their maximum extent poleward while making each interglacial unique based upon the different tree and shrubs species that dominant during each specific period. Ice sheets in the Arctic, such as the Greenland Ice Sheet, were significantly reduced during some interglacials, allowing summer temperatures to increase almost 2 to 5 °C warmer than present. The observed response of the region's climate and terrestrial vegetation distribution to super-interglacial forcing is still not fully understood and creates a challenge for climate modeling and the study of Arctic amplification. Such examples are the extreme warmth at MIS 11c despite lower than modern GHG concentrations and the muted response in proxy records for MIS 5e, despite extreme summer insolation intensity. Additionally, modeling studies showed overall drier conditions in the earlier interglacials (11c and 31) relative to pollen analysis.
The significant warming in the circum-Arctic can be linked to major deglaciation events in Antarctica, demonstrating possible intrahemispheric linkages between the Arctic and Antarctic climate on glacial-interglacial timescales.
Large northern hemisphere ice sheets during major glaciation events can be linked to Arctic aridification and extremely cold annual temperatures. The combination of 78 increased Arctic sea ice and increased surface albedo allows the Arctic to significantly cool and dry out during these events. This is demonstrated in the Lake El'gygytgyn core by multiproxy analyses and a transition to shrub vegetation due to the lack of precipitation. The climate modeling showed here suggests extreme Arctic aridification after 2.7 Ma was a consequence of the episodic expansion of ice sheets, which affected dominant atmospheric pressure patterns, the storm track and a general southward shift of precipitation in the Beringian sector of the Arctic. 79